阅读说明：本技术 主动降噪方法、装置、系统以及相关设备 (Active noise reduction method, device and system and related equipment ) 是由 方泽凯 朱嘉俊 郝鑫 于 2020-04-27 设计创作，主要内容包括：本发明涉及主动降噪方法及相关设备,通过在误差噪声的通路上引入补偿滤波器以补偿FXLMS算法中引入次级通路频响后造成的衰减,使得经过滤波后的误差噪声再进行LMS运算后更新自适应滤波器时能够获得更好滤波系数,减低次级通路频响衰减对降噪性能的影响,使整体频段上的降噪性能更为均衡,进一步提高降噪性能。(The invention relates to an active noise reduction method and related equipment, wherein a compensation filter is introduced into an error noise path to compensate attenuation caused by introducing a secondary path frequency response into an FXLMS algorithm, so that better filter coefficients can be obtained when an adaptive filter is updated after LMS operation is carried out on filtered error noise, the influence of the secondary path frequency response attenuation on the noise reduction performance is reduced, the noise reduction performance on the whole frequency band is more balanced, and the noise reduction performance is further improved.)

1. An active noise reduction method, comprising the steps of:

s01, picking up noise source noise through a reference microphone to obtain a reference noise signal x (n);

s02, making the reference noise signal x (n) pass through an adaptive filter to generate a noise-eliminating driving signal, and driving a noise-eliminating loudspeaker to generate secondary noise y (n) through a secondary path;

s03, picking up the superposed signal of the primary noise d (n) and the secondary noise y (n) through an error microphone, and obtaining an error noise signal e (n) at the current moment; wherein, the primary noise d (n) is a signal formed by the noise of the noise source transmitted to a sound receiving position through a primary path;

s04, carrying out X-filtering processing on the reference noise signal X (n) at the current moment by an X-filter to obtain a corrected reference noise signal X' (n); wherein the X-filter is an equivalent filter of the secondary path, and a transfer function of the X-filter is an estimation value of a transfer function of the secondary path;

s05, the error noise signal e (n) at the current moment is subjected to compensation filtering processing by a compensation filter to obtain a corrected error noise signal e' (n); the compensation filter is used for attenuation compensation of the X-filter, and the compensation filter performs gain compensation on the error noise signal e (n) in a preset frequency band attenuated by the X-filter;

s06, updating the coefficient of the adaptive filter at the next time based on the corrected reference noise signal x '(n) and the corrected error noise signal e' (n);

and S07, repeating the steps S01-S06 until the mean square value of the error noise signal e (n) converges to a preset value, wherein the secondary noise y (n) is a sound wave with the same amplitude and opposite phase with the primary noise d (n), and the secondary noise y (n) can counteract the primary noise d (n).

2. The active noise reduction method according to claim 1, wherein after step S04 and before step S06, the method further comprises:

and S04 ', the corrected reference noise signal x ' (n) is subjected to compensation filtering processing by the compensation filter, and the updated corrected reference noise signal x ' (n) is obtained.

3. The active noise reduction method according to claim 2, wherein before step S04 ″, the method further comprises:

s04 ', the corrected reference noise signal x ' (n) is shaped and filtered by a shaping filter, and the updated corrected reference noise signal x ' (n) is obtained;

correspondingly, before step S05, the method further includes:

and S05', shaping and filtering the error noise signal e (n) at the current moment by a shaping filter to obtain the updated error noise signal e (n) at the current moment, wherein the frequency response function of the shaping filter is a curve matched with the auditory curve of the human ear.

4. Active noise reduction method according to any of claims 1-3, characterized in that the frequency response of the compensation filter is the inverse of the X-filter frequency response or the inverse of the secondary path frequency response.

5. The active noise reduction method according to any of claims 1-3, wherein the frequency response of the compensation filter is a ratio of the primary path frequency response and the secondary path frequency response.

6. The active noise reduction method according to claim 5, wherein the coefficients of the compensation filter are obtained from N iterations of the coefficients of the adaptive filter, where N is an integer.

7. The active noise reduction method according to claim 6, wherein the calculation of the coefficients of the compensation filter comprises the following steps:

step S201, keeping the coefficient of the compensation filter unchanged in T time, iteratively updating the coefficient of the adaptive filter through steps S01-S06, and updating the coefficient of the compensation filter with the coefficient at T time;

step S202, executing step S201 for N times, and obtaining the coefficient of the compensation filter after iterative updating;

wherein T is a preset convergence time.

8. An active noise reduction device, comprising:

the reference noise pickup module is used for picking up noise of the noise source through a reference microphone to obtain a reference noise signal x (n);

a secondary noise generation module, for making the reference noise signal x (n) pass through an adaptive filter to generate a noise cancellation driving signal, and the noise cancellation driving signal drives a noise cancellation loudspeaker to generate secondary noise y (n) through a secondary path;

the LMS module is used for iteratively updating the coefficient of the self-adaptive filter until the mean square value of the error noise signal converges to a preset value;

when the mean square value of the error noise signal e (n) converges to the preset value, the secondary noise y (n) generated by the secondary noise generation module is a sound wave with the same amplitude and opposite phase with the primary noise d (n), and the secondary noise y (n) can be counteracted with the primary noise d (n);

wherein, the LMS module includes:

an error noise signal obtaining unit, configured to pick up a signal obtained by superimposing the primary noise d (n) and the secondary noise y (n) by using an error microphone, and obtain an error noise signal e (n) at the current time; wherein, the primary noise d (n) is a signal formed by the noise of the noise source transmitted to a sound receiving position through a primary path;

the reference noise correction unit is used for carrying out X-filtering processing on the reference noise signal X (n) at the current moment through an X-filter to obtain a corrected reference noise signal X' (n); wherein the X-filter is an equivalent filter of the secondary path, and a transfer function of the X-filter is an estimation value of a transfer function of the secondary path;

the error noise correction unit is used for performing compensation filtering processing on the error noise signal e (n) at the current moment through a compensation filter to obtain a corrected error noise signal e' (n); the compensation filter is used for attenuation compensation of the X-filter, and the compensation filter performs gain compensation on the error noise signal e (n) in a preset frequency band attenuated by the X-filter;

a filter coefficient updating unit configured to update a coefficient of the adaptive filter at a next time based on the corrected reference noise signal x '(n) and the corrected error noise signal e' (n).

9. The active noise reduction device according to claim 8, wherein the reference noise modification unit is further configured to perform the following processing after performing the X-filtering processing:

and performing compensation filtering processing on the corrected reference noise signal x '(n) through the compensation filter to obtain an updated corrected reference noise signal x' (n).

10. The active noise reduction device according to claim 9, wherein the reference noise modification unit is further configured to perform the following processing before performing the filtering compensation processing:

s04 ', the corrected reference noise signal x ' (n) is shaped and filtered by a shaping filter, and the updated corrected reference noise signal x ' (n) is obtained;

correspondingly, the error noise correction unit is further configured to, before performing the compensation filtering process, perform the following processes:

and shaping and filtering the error noise signal e (n) at the current moment by a shaping filter to obtain an updated error noise signal e (n) at the current moment, wherein the frequency response function of the shaping filter is a curve matched with an auditory curve of a human ear.

11. The active noise reduction device of any of claims 8-10, wherein the frequency response of the compensation filter is an inverse of the X-filter frequency response or an inverse of the secondary path frequency response.

12. The active noise reduction device of any of claims 8-10, wherein the frequency response of the compensation filter is a ratio of the primary path frequency response and the secondary path frequency response.

13. The active noise reduction device according to claim 12, further comprising a compensation coefficient calculation module for obtaining the coefficients of the compensation filter through N iterations of the coefficients of the adaptive filter; wherein N is an integer.

14. The active noise reduction device of claim 13, wherein the compensation coefficient calculation comprises:

the adaptive iteration unit is used for keeping the coefficient of the compensation filter unchanged in T time, iteratively updating the coefficient of the adaptive filter through an LMS module, and updating the coefficient of the compensation filter by the coefficient at T moment;

the compensation iteration unit is used for obtaining the coefficient of the compensation filter after the adaptive iteration unit carries out updating for N times;

wherein T is a preset convergence time.

15. An active noise reduction apparatus for performing the active noise reduction method of any one of claims 1-7, comprising: an adaptive filter, an X-filter, a compensation filter, and a coefficient update module, wherein,

the input end of the adaptive filter is connected with a reference microphone to receive a reference noise signal x (n), and the output end of the adaptive filter outputs secondary noise y (n) to a noise elimination loudspeaker;

the input end of the X-filter is connected with the reference microphone to receive a reference noise signal X (n), and the output end of the X-filter is connected with one input end of the coefficient updating module;

the input end of the compensation filter is connected with an error microphone to obtain an error noise signal e (n), and the output end of the compensation filter is connected with one input end of the coefficient updating module;

and after the coefficient updating module calculates the coefficient of the self-adaptive filter, the coefficient is output to the self-adaptive filter to update the coefficient of the self-adaptive filter.

16. An active noise reduction system comprising a reference microphone for picking up noise source noise, an error microphone for picking up an error noise signal and an active noise reduction control device according to any one of claims 8-15.

17. A headset having an earmuff, further having the active noise reduction system of claim 16, wherein the reference microphone is configured to pick up a noise signal external to the earmuff and the error microphone is configured to pick up a noise signal internal to the earmuff.

18. Audio device, characterized in that it comprises an active noise reduction means according to any of claims 8-15.

19. A chip having an integrated circuit thereon, characterized in that the integrated circuit is designed for implementing the method as claimed in any of claims 1-7.

20. A storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, performs the method according to any one of claims 1-7.

Technical Field

The present invention relates to the field of active noise reduction, and in particular, to an active noise reduction method, apparatus, system, headphone, audio device, chip, and storage medium.

Background

Passive Noise reduction (ANC) and Active Noise Control (ANC) are the two main Noise cancellation methods at present. The principle of the relatively common passive noise reduction methods, such as using earplugs, earmuffs, etc., is to eliminate noise energy by using acoustic materials or acoustic structures, such as sound insulation and absorption, etc., and the control of high-frequency noise is usually very effective. For low-frequency noise, passive noise reduction requires increasing the weight of a control material or the thickness of a sound-absorbing material to obtain a good noise reduction effect, so that a passive noise reduction member is heavy and bulky. ANC is based on the principle of interference, and generates a waveform with the same amplitude and opposite phase as the noise through a secondary path (a noise cancellation speaker, etc.) for cancellation. Compared with passive noise reduction, the active noise reduction low-frequency performance is good, and the original structure does not need to be adjusted greatly.

Existing ANC systems typically employ both feed-forward and feed-back configurations. The feedforward structure uses two microphones to receive external noise, the reference microphone receives external noise source signal, after the adaptive filter generates driving signal, the noise-canceling loudspeaker is driven to output sound wave with opposite phase to the noise, then the error microphone receives residual noise signal, as feedback parameter, the coefficient of the adaptive filter is adjusted until reaching the desired value or the noise is eliminated, as shown in fig. 1. At present, for the adaptive filter, the coefficients are usually obtained by using a conventional adaptive algorithm such as LMS (least mean square error criterion), i.e. the variance of the error noise is minimized by adjusting the adaptive filter coefficients, as shown in fig. 2, which is an ANC structure diagram of FXLMS algorithm modified based on LMS algorithm, wherein the primary path refers to the path from the reference microphone position to the error microphone position and has a transfer function of the primary path frequency response, i.e. the system frequency response p (z), and the secondary path refers to the path from the filter output to the error microphone via the noise cancellation speaker and has a transfer function of the secondary path frequency response s (z). In the active noise reduction process, the FXLMS algorithm considers the delay problem caused by the existence of the secondary path, and therefore, a filter (generally referred to as X-filter) equivalent to the transfer function of the secondary path is added to filter the reference signal, so as to eliminate the stability problem of the system, that is, equivalently, the secondary path frequency response is introduced into the ANC system, but the FXLMS algorithm does not consider the influence of the introduced secondary path frequency response on the final noise reduction performance, as shown in fig. 3, the secondary path frequency response is obtained based on the simulation of the FXLMS algorithm, the frequency response is obtained by collecting data in the real environment, as shown in fig. 4, the noise reduction performance is obtained by using the conventional FXLMS algorithm, and it can be seen from the noise reduction effect of fig. 4 that the noise reduction effect of the active noise reduction performance (mainly embodied in the power of the error signal) based on the FXLMS algorithm in different frequency bands is influenced by the secondary path frequency response, when the frequency response of the secondary channel is attenuated, the noise reduction effect is poor. Therefore, the influence of the frequency response of the introduced secondary path equivalent filter on the noise reduction effect needs to be further solved.

Disclosure of Invention

Based on the above situation, the main objective of the present invention is to provide an active noise reduction method, apparatus, system, headphone audio device, chip and storage medium, in which a compensation filter is added to a secondary path to compensate the influence on the overall noise reduction performance caused by introducing the frequency response of the secondary path in the prior art, so as to further optimize the noise reduction performance.

In order to achieve the purpose, the technical scheme adopted by the invention is as follows:

an active noise reduction method comprising the steps of:

s01, picking up noise source noise through a reference microphone to obtain a reference noise signal;

s02, making the reference noise signal x (n) pass through an adaptive filter to generate a noise-eliminating driving signal, and driving a noise-eliminating loudspeaker to generate secondary noise y (n) through a secondary path;

s03, picking up the superposed signal of the primary noise d (n) and the secondary noise y (n) through an error microphone, and obtaining an error noise signal e (n) at the current moment; wherein, the primary noise d (n) is a signal formed by the noise of the noise source transmitted to a sound receiving position through a primary path;

s04, carrying out X-filtering processing on the reference noise signal X (n) at the current moment by an X-filter to obtain a corrected reference noise signal X' (n); wherein the X-filter is an equivalent filter of the secondary path, and a transfer function of the X-filter is an estimation value of a transfer function of the secondary path;

s05, the error noise signal e (n) at the current moment is subjected to compensation filtering processing by a compensation filter to obtain a corrected error noise signal e' (n); the compensation filter is used for attenuation compensation of the X-filter, and the compensation filter performs gain compensation on the error noise signal e (n) in a preset frequency band attenuated by the X-filter;

s06, updating the coefficient of the adaptive filter at the next time based on the corrected reference noise signal x '(n) and the corrected error noise signal e' (n);

and S07, repeating the steps S01-S06 until the mean square value of the error noise signal e (n) converges to a preset value, wherein the secondary noise y (n) is a sound wave with the same amplitude and opposite phase with the primary noise d (n), and the secondary noise y (n) can counteract the primary noise d (n).

Preferably, after step S04 and before step S06, the method further comprises:

and S04 ', the corrected reference noise signal x ' (n) is subjected to compensation filtering processing by the compensation filter, and the updated corrected reference noise signal x ' (n) is obtained.

Preferably, before step S04 ″, the method further comprises: s04 ', the corrected reference noise signal x ' (n) is shaped and filtered by a shaping filter, and the updated corrected reference noise signal x ' (n) is obtained; correspondingly, before step S05, the method further includes: and S05', shaping and filtering the error noise signal e (n) at the current moment by a shaping filter to obtain the updated error noise signal e (n) at the current moment, wherein the frequency response function of the shaping filter is a curve matched with the auditory curve of the human ear.

Preferably, the frequency response of the compensation filter is the inverse of the X-filter frequency response or the inverse of the secondary path frequency response.

Preferably, the frequency response of the compensation filter is a ratio of the primary path frequency response and the secondary path frequency response.

Preferably, the coefficients of the compensation filter are obtained from N iterations of the coefficients of the adaptive filter, where N is an integer.

Preferably, the calculation process of the coefficients of the compensation filter includes the steps of:

step S201, keeping the coefficient of the compensation filter unchanged in T time, iteratively updating the coefficient of the adaptive filter through steps S01-S06, and updating the coefficient of the compensation filter with the coefficient at T time;

step S202, executing step S201 for N times, and obtaining the coefficient of the compensation filter after iterative updating;

wherein T is a preset convergence time.

To achieve the above object, the present invention further provides an active noise reduction device, including:

the reference noise pickup module is used for continuously picking up noise source noise through a reference microphone to obtain a reference noise signal x (n), and transmitting the noise source noise to a sound reception position through a primary path to form primary noise d (n);

a secondary noise generation module, for making the reference noise signal x (n) pass through an adaptive filter to generate a noise cancellation driving signal, and the noise cancellation driving signal drives a noise cancellation loudspeaker to generate secondary noise y (n) through a secondary path;

the LMS module is used for iteratively updating the coefficient of the self-adaptive filter until the mean square value of the error noise signal converges to a preset value;

when the mean square value of the error noise signal e (n) converges to the preset value, the secondary noise y (n) generated by the secondary noise generation module is a sound wave with the same amplitude and opposite phase with the primary noise d (n), and the secondary noise y (n) can be counteracted with the primary noise d (n);

wherein, the LMS module includes:

an error noise signal obtaining unit, configured to pick up a signal obtained by superimposing the primary noise d (n) and the secondary noise y (n) by using an error microphone, and obtain an error noise signal e (n) at the current time; wherein, the primary noise d (n) is a signal formed by the noise of the noise source transmitted to a sound receiving position through a primary path;

the reference noise correction unit is used for carrying out X-filtering processing on the reference noise signal X (n) at the current moment through an X-filter to obtain a corrected reference noise signal X' (n); wherein the X-filter is an equivalent filter of the secondary path, and a transfer function of the X-filter is an estimation value of a transfer function of the secondary path;

the error noise correction unit is used for performing compensation filtering processing on the error noise signal e (n) at the current moment through a compensation filter to obtain a corrected error noise signal e' (n); the compensation filter is used for attenuation compensation of the X-filter, and the compensation filter performs gain compensation on the error noise signal e (n) in a preset frequency band attenuated by the X-filter;

a filter coefficient updating unit configured to update a coefficient of the adaptive filter at a next time based on the corrected reference noise signal x '(n) and the corrected error noise signal e' (n).

To achieve the above object, the present invention further provides an active noise reduction apparatus for performing the active noise reduction method described above, including: the noise reduction device comprises an adaptive filter, an X-filter, a compensation filter and a coefficient updating module, wherein the input end of the adaptive filter is connected with a reference microphone to receive a reference noise signal X (n), and the output end of the adaptive filter outputs secondary noise y (n) to a noise reduction loudspeaker; the input end of the X-filter is connected with the reference microphone to receive a reference noise signal X (n), and the output end of the X-filter is connected with one input end of the coefficient updating module; the input end of the compensation filter is connected with an error microphone to obtain an error noise signal e (n), and the output end of the compensation filter is connected with one input end of the coefficient updating module; and after the coefficient updating module calculates the coefficient of the self-adaptive filter, the coefficient is output to the self-adaptive filter to update the coefficient of the self-adaptive filter.

To achieve the above object, the present invention further provides an active noise reduction system, which includes a reference microphone for picking up noise of a noise source, an error microphone for picking up an error noise signal, and the active noise reduction control apparatus as described above.

To achieve the above object, the present invention further provides a headset having an ear cup and further having an active noise reduction system as described above, wherein the reference microphone is arranged to pick up a noise signal outside the ear cup and the error microphone is arranged to pick up a noise signal inside the ear cup.

To achieve the above object, the present invention further provides an audio device comprising the active noise reduction apparatus as described above.

To achieve the above object, the present invention also provides a chip having an integrated circuit thereon, the integrated circuit being designed to implement the active noise reduction method as described above.

To achieve the above object, the present invention further provides a storage medium storing a computer program, which when executed by a processor, executes the active noise reduction method as described above.

Has the advantages that:

in the embodiment of the invention, the compensation filter is introduced into the path of the error noise to compensate the attenuation caused by introducing the frequency response of the secondary path into the FXLMS algorithm, so that better filter coefficients can be obtained when the filtered error noise is subjected to LMS operation and then the adaptive filter is updated, the influence of the attenuation of the frequency response of the secondary path on the noise reduction performance is reduced, the noise reduction performance on the whole frequency band is more balanced, and the noise reduction performance is further improved.

Other advantages of the present invention will be described in the detailed description, and those skilled in the art will understand the technical features and technical solutions presented in the description.

Drawings

Preferred embodiments according to the present invention will be described below with reference to the accompanying drawings. In the figure:

FIG. 1 is a diagram of an application environment of an ANC system in the prior art;

FIG. 2 is a schematic structural diagram of an ANC system based on FXLMS algorithm in the prior art;

FIG. 3 is a schematic diagram illustrating a simulation of a secondary path frequency response of an ANC system based on the FXLMS algorithm in the prior art;

FIG. 4 is a schematic diagram illustrating simulation of active noise reduction performance of the ANC system corresponding to FIG. 3;

FIG. 5 is a schematic flow chart illustrating an active noise reduction method according to an embodiment of the present invention;

FIG. 6 is a schematic diagram of an algorithm framework for an active noise reduction algorithm according to an embodiment of the present invention;

FIG. 7 is a schematic diagram of an algorithm framework for an active noise reduction algorithm in another embodiment of the present invention;

FIG. 8 is a schematic diagram of an algorithm framework for an active noise reduction algorithm in accordance with another embodiment of the present invention;

FIG. 9 shows a shaping filter H according to an embodiment of the invention_nw(z) a frequency response simulation diagram;

FIG. 10 is a schematic diagram of an algorithm framework for an active noise reduction algorithm in accordance with another embodiment of the present invention;

fig. 11 is a comparison graph of the active noise reduction performed by the conventional FXLMS algorithm and the noise reduction performed by the improved active noise reduction method according to the embodiment of the present invention;

FIG. 12 is a comparison chart of the convergence process of the error noise signals after the active noise reduction is performed by using the conventional LMS algorithm and the noise reduction is performed by using the improved active noise reduction method of the present embodiment;

FIG. 13 is a functional block diagram of an active noise reduction apparatus according to an embodiment of the present invention;

FIG. 14 is a schematic structural diagram of an active noise reduction system according to an embodiment of the present invention;

fig. 15 is a schematic structural diagram of an active noise reduction system according to another embodiment of the present invention.

Detailed Description

In order to describe the technical solutions of the present invention in more detail to facilitate further understanding of the present invention, the following describes specific embodiments of the present invention with reference to the accompanying drawings. It should be understood, however, that all of the illustrative embodiments and descriptions thereof are intended to illustrate the invention and are not to be construed as the only limitations of the invention.

Referring to fig. 5, a schematic flow chart of an active noise reduction method according to an embodiment of the invention is shown. In this embodiment, the active noise reduction method includes the following steps:

s01, picking up noise source noise through a reference microphone to obtain a reference noise signal x (n);

s02, making the reference noise signal x (n) pass through an adaptive filter to generate a noise-eliminating driving signal, and driving a noise-eliminating loudspeaker to generate secondary noise y (n) through a secondary path;

s03, picking up the superposed signal of the primary noise d (n) and the secondary noise y (n) through an error microphone, and obtaining an error noise signal e (n) at the current moment; wherein, the primary noise d (n) is a signal formed by the noise of the noise source transmitted to a sound receiving position through a primary path;

s04, carrying out X-filtering processing on the reference noise signal X (n) at the current moment by an X-filter to obtain a corrected reference noise signal X' (n); wherein the X-filter is an equivalent filter of the secondary path, and a transfer function of the X-filter is an estimation value of a transfer function of the secondary path;

s05, the error noise signal e (n) at the current moment is subjected to compensation filtering processing by a compensation filter to obtain a corrected error noise signal e' (n); the compensation filter is used for attenuation compensation of the X-filter, and the compensation filter performs gain compensation on the error noise signal e (n) in a preset frequency band attenuated by the X-filter;

s06, updating the coefficient of the adaptive filter at the next time based on the corrected reference noise signal x '(n) and the corrected error noise signal e' (n);

and S07, repeating the steps S01-S06 until the mean square value of the error noise signal e (n) converges to a preset value, wherein the secondary noise y (n) is a sound wave with the same amplitude and opposite phase with the primary noise d (n), and the secondary noise y (n) can counteract the primary noise d (n).

Please refer to fig. 6, which is a schematic diagram of an algorithm framework of the FXLMS algorithm after the improvement in the active noise reduction method in this embodiment. In the present embodiment, the compensation filter H_s(z) for compensating the attenuation of the frequency response of the X-filter, the reference noise signal X (n) arriving at the ears of the person via the primary path p (z) forming primary noise d (n), the secondary noise y (n) for cancelling the primary noise d (n) being generated by a noise canceling loudspeaker whose noise canceling drive signal y'(n) the reference noise signal x (n) is generated by the adaptive filter W (z), the noise-canceling driving signal y' (n) is generated after passing through the adaptive filter W (z), the secondary noise y (n) reaching the human ear is generated by the noise-canceling loudspeaker through the secondary path S (z), and the ideal secondary noise y (n) and the primary noise d (n) are waveforms with the same amplitude and opposite phases, so that the primary noise d (n) can be cancelled, and the noise-canceling effect is achieved. In practice, the secondary noise y (n) has a process of gradually approaching the primary noise d (n), which is realized by iteratively adjusting the coefficients of the adaptive filter w (z) for a plurality of times according to the error noise signals e (n) of the two, and finally, when the adaptive filter reaches a convergence state, the mean square value of the error noise signals e (n) converges to a preset value. The adjustment of the coefficients of the adaptive filter w (z) is performed by the LMS algorithm block, which has two inputs, respectively from the reference noise signal x (n) and the error noise signal e (n). The FXLMS algorithm is mainly modified for a reference noise signal X (n), and an X-filter S' (z) is added to filter the reference noise signal X (n) so as to compensate the problem that the input on two sides of the LMS is not synchronous due to the existence of a secondary path S (z). In this embodiment, the effect of the introduced attenuation of the frequency response of the X-filter S ' (z) on the noise reduction performance is further compensated mainly by modifying the other input error noise signal e (n), as shown in fig. 3 and 4, the attenuation of the error noise in the attenuation band of the frequency response of the X-filter S ' (z), especially in the initial part of 0HZ and the part after 1.5KH, causes the power of the error noise signal e (n) to approach the power before noise reduction, so that, in order to compensate the attenuation band of the X-filter S ' (z), a cascade compensation filter H (H) may be connected on the path where the error microphone is located_s(z) performing gain compensation of the same amplitude on the error noise signal e (n) in the same frequency band. Since the X-filter S' (z) is the equivalent filter of the secondary path, the compensation filter H_sThe frequency response curve of (z) may adopt a curve which varies inversely with the frequency response curve of the secondary path, so that the error noise signal e (n) may be filter-shaped in the direction opposite to the frequency response curve of the secondary path, thereby realizing compensation for the introduced attenuation of the frequency response of the secondary path.

It is to be understood that fig. 3 and 4 are only used to help explain the effect of introducing the X-filter S' (z) on the noise reduction performance, but are not to be considered as limiting the compensation band. The frequency band that is actually attenuated for different noise and different systems may be any frequency band.

It is understood that in a preferred embodiment, the compensation filter H may not process the frequency band with the X-filter S' (z) corresponding to the error noise signal e (n) with good noise reduction performance_s(z) all-pass processing the error noise signal e (n).

In the prior art, S' (z) can be generally obtained in an online or offline manner, and the present invention will not be described in detail.

In this embodiment, based on the modified FXLMS algorithm, the update process of the adaptive filter w (z) coefficient is as follows:

w(n+1)＝w(n)+ue′(n)*x′(n)；

e′(n)＝e(n)*h_s(n)；

x(n)′＝x(n)*s′(n)；

wherein w (n) represents a coefficient vector of the adaptive filter w (z); e' (n) is the corrected output of the error noise signal e (n); s' (n) is the impulse response of the X-filter; h is_s(n) is a compensation filter H_s(z) impact response.

In the embodiment of the invention, the compensation filter is introduced into the path of the error noise to compensate the attenuation caused by introducing the frequency response of the secondary path into the FXLMS algorithm, so that better filter coefficients can be obtained when the filtered error noise is subjected to LMS operation and then the adaptive filter is updated, the influence of the attenuation of the frequency response of the secondary path on the noise reduction performance is reduced, the noise reduction performance on the whole frequency band is more balanced, and the noise reduction performance is further improved.

Preferably, in an optional embodiment, in order to further improve the system stability and eliminate the system instability possibly caused by the delay caused by introducing a new filter, a compensation filter H may be further cascaded on the path where the other input end of the LMS module, i.e. the reference noise signal x (n), is located_s(z). At this time, after the step S04 and before the step S06, the method further includes:

and S04 ', the corrected reference noise signal x ' (n) is subjected to compensation filtering processing by the compensation filter, and the updated corrected reference noise signal x ' (n) is obtained.

As shown in fig. 7, the update procedure of the adaptive filter w (z) coefficient is as follows:

w(n+1)＝w(n)+ue′(n)*x′(n)；

e′(n)＝e(n)*h_s(n)；

x(n)′＝x(n)*s′(n)*h_s(n)；

wherein w (n) represents a coefficient vector of the adaptive filter w (z); e' (n) is the output of the error noise signal e (n) after two times of correction; s' (n) is the impulse response of the X-filter; h is_s(n) is a compensation filter H_s(z) impact response.

Preferably, in an alternative embodiment, the two input quantities of the LMS may be shaped and filtered simultaneously, so that the filtered input quantities can match the auditory curve of human ears, in which case, before step S04 ″, the method further includes:

s04 ', the corrected reference noise signal x ' (n) is shaped and filtered by a shaping filter, and the updated corrected reference noise signal x ' (n) is obtained;

correspondingly, before step S05, the method further includes:

s05', the error noise signal e (n) at the current time is shaped and filtered by a shaping filter, so as to obtain an updated error noise signal e (n) at the current time.

Referring to fig. 8 synchronously, considering the fact that the human ear experiences inconsistency at different frequencies subjectively, i.e., the loudness of the human ear experiences signals of equal amplitude at different frequencies is inconsistent, the shaping filter H_nwThe frequency response function of (z) is a curve that matches the auditory curve of the human ear. Adding a shaping filter H_nw(z) filtering and shaping the error noise signal e (n) and the reference noise signal according to a mode opposite to a psychoacoustic curve of the human ear, inputting the filtered error noise signal e (n) and the reference noise signal into the LMS module to adjust a coefficient vector of the adaptive filter W (z), and considering the human ear on different frequencies in the process of generating the offset noiseThe loudness of signals with the same amplitude is inconsistent, and better filter coefficients can be obtained.

It is understood that, in the present embodiment, the error noise signal e (n) and the reference noise signal x (n) are filtered for a plurality of times, and then input to the LMS module for coefficient update calculation. The filtering order is not intended to limit the embodiments of the present invention. In different real-time scenarios, the filtering order of the error noise signal e (n) and the reference noise signal x (n) can be adjusted at will. For example, for the reference noise signal X (n), compensation filtering, shaping filtering, and X-filtering may be performed.

Preferably, in an alternative embodiment, the shaping filter H_nw(z) impact response h_nw(n) A weighting curve can be selected, in which case the shaping filter H_nw(z) is a weighting filter. The a weighted filter is generated by the following equation:

A＝2.0+20log(R_a(f))；

shaping filter H_nwThe frequency response of (z) is shown in fig. 9, whereby a shaping filter H is added_nw(z) filtering and shaping the error noise signal e '(n) according to a mode opposite to a human ear psychoacoustic curve, inputting the corrected error noise signal e' (n) after filtering into an LMS module, adjusting a coefficient vector of an adaptive filter W (z), considering that loudness of human ears on signals with the same amplitude on different frequencies is inconsistent in the process of generating offset noise, and shaping the noise by matching the human ear psychoacoustic curve to obtain noise reduction performance according with psychoacoustic, so that the noise reduction experience of a user is further improved.

Preferably, in an alternative embodiment, the compensation filter H_sFrequency response H of (z)_s(z) is the inverse of the X-filter frequency response S' (z) or the inverse of the secondary path frequency response. Due to the compensation filter H_s(z) is for reducing the effect of the attenuation of the secondary path frequency response introduced by the X-filter frequency response S' (z), since the X-filter is a secondary filterThe estimated values of the stage paths, S' (z) and S (z), can both be viewed as the same curve, thus, in one embodiment, the compensation filter H_sFrequency response H of (z)_s(z) is the inverse of the X-filter frequency response S' (z) or the inverse of the secondary path frequency response S (z), so that the effect of the secondary path frequency response on the noise reduction performance, i.e. h, is completely cancelled_sThe ideal value of (n) is the estimate of the inverse of s' (n) or s (n), in this case. The algorithm framework shown in fig. 8 has been practically optimized as shown in fig. 9, when the compensation filter H is in the path of the reference noise signal x (n)_s(z) and the X-filter S' (z) practically cancel each other out, only the delay problem needs to be considered to align the reference noise signal X (n) and the error noise signal e (n), in this case, d (z) represents the delay operation, and the update process of the coefficients of the adaptive filter w (z) is as follows:

w(n+1)＝w(n)+ue′(n)*x′(n)；

e′(n)＝e(n)*h_nw(n)*h_s(n)；

x(n)′＝x(n)*h_nw(n)*delay(n)；

delay (n) delays the reference noise signal x (n) to avoid the two inputs being asynchronous.

Preferably, in an alternative embodiment, the frequency response H of the compensation filter_s(z) is the ratio of the primary path to the secondary path frequency response, i.e.

In practical engineering applications, the inverse of the secondary path frequency response, or the inverse of the X-filter, is not easily obtained. In the FXLMS algorithm, when the adaptive filter W (z) converges, the coefficient frequency response of the adaptive filter

In typical active noise reduction systems, p (z) exhibits a low-pass response, and therefore, w (z), which is close to the inverse of the secondary path frequency response, may be used instead of the inverse of the secondary path frequency response, in which case,

preferably, in an alternative embodiment, the compensation filter H may be determined in an embodiment by successive approximation in an iterative manner_s(z) coefficient of the compensation filter H_s(z) the coefficients of the adaptive filter w (z) are obtained over N iterations, the calculation process comprising the steps of:

a step S201 of iteratively updating the coefficient of the adaptive filter through steps S01-S06 based on the corrected reference noise signal x '(n) and the corrected error noise signal e' (n) while keeping the coefficient of the compensation filter unchanged for a time T, and updating the coefficient of the compensation filter with the coefficient at the time T;

step S202, executing step S201 for N times, and obtaining the coefficient of the compensation filter after iterative updating;

compensation filter H_sThe coefficients of (z) are related to those of the adaptive filter w (z), but the time at which the adaptive filter w (z) is used is optional, and in this case, the time can be determined in an iterative manner, and the number of iterations can be determined through a plurality of experiments, so that the final error noise signal e (n) can be converged within a required time. For FIR, the impulse response is also the filter coefficients. In the initial state, the first iteration, the compensation filter H_sThe coefficient of (z) can be set to 1, in which case the frequency response attenuation of the compensating sub-path, i.e. h, is not actually taken into account_s(n) 1, compensation filter H_sThe iterative process of the coefficients of (z) is as follows:

in a specific implementation scenario, the iteration number N and the convergence time T may be adjusted accordingly according to an actual system, and the time T ensures that the error noise signal e (N) converges below a preset value in the adaptive process of the filter. Please refer to fig. 11 and 12 at the same time, fig. 11 is a comparison graph of active noise reduction by using a conventional FXLMS algorithm and noise reduction effect by using the active noise reduction method improved in the present embodiment, and fig. 12 is a comparison graph of convergence processes of error noise signals of the two, as is apparent from the graphs, the active noise reduction method of the present embodiment has better noise reduction performance than the conventional algorithm, obviously reduces noise power in the entire frequency band, and has better error noise signal convergence effect, and based on simulation data calculation in the graph, the active noise reduction method of the present embodiment reduces the noise power in the total frequency band from the conventional-8 dB to about-21 dB, has balanced overall noise reduction effect, and has higher reference significance in engineering practice.

The invention further provides an active noise reduction device. Referring to fig. 13, a functional block diagram of an active noise reduction device 10 according to an embodiment is shown. In the present embodiment, the active noise reduction device 10 includes:

a reference noise pickup module 11, configured to continuously pick up noise source noise through a reference microphone to obtain a reference noise signal;

a secondary noise generation module 12, configured to make the reference noise signal x (n) pass through an adaptive filter to generate a noise cancellation driving signal, and the noise cancellation speaker is driven by a secondary path to generate secondary noise y (n);

the LMS module 13 is configured to iteratively update the coefficient of the adaptive filter until the mean square value of the error noise signal converges to a preset value;

the LMS module 13 includes:

an error noise signal obtaining unit 131, configured to pick up a signal obtained by superimposing the primary noise d (n) and the secondary noise y (n) through an error microphone, and obtain an error noise signal e (n) at the current time; wherein, the primary noise d (n) is a signal formed by the noise of the noise source transmitted to a sound receiving position through a primary path;

a reference noise modification unit 132, configured to perform X-filtering processing on the reference noise signal X (n) at the current time through an X-filter to obtain a modified reference noise signal X' (n); wherein the X-filter is an equivalent filter of the secondary path, and a transfer function of the X-filter is an estimation value of a transfer function of the secondary path;

an error noise correction unit 133, configured to perform compensation filtering processing on the error noise signal e (n) at the current time through a compensation filter to obtain a corrected error noise signal e' (n); the compensation filter is used for attenuation compensation of the X-filter, and the compensation filter performs gain compensation on the error noise signal e (n) in a preset frequency band attenuated by the X-filter;

a filter coefficient updating unit 134, configured to update the coefficient of the adaptive filter at the next time based on the modified reference noise signal x '(n) and the modified error noise signal e' (n).

Preferably, in an alternative embodiment, the reference noise modification unit is further configured to, after performing the X-filtering process, perform the following processes:

and performing compensation filtering processing on the corrected reference noise signal x '(n) through the compensation filter to obtain an updated corrected reference noise signal x' (n).

Preferably, in an optional embodiment, the reference noise modification unit is further configured to, before performing the filtering compensation process, perform the following processes:

shaping and filtering the corrected reference noise signal x '(n) through a shaping filter to obtain an updated corrected reference noise signal x' (n);

correspondingly, the error noise correction unit is further configured to, before performing the compensation filtering process, perform the following processes:

and shaping and filtering the error noise signal e (n) at the current moment by a shaping filter to obtain an updated error noise signal e (n) at the current moment, wherein the frequency response function of the shaping filter is a curve matched with an auditory curve of a human ear.

Preferably, in an alternative embodiment, the frequency response of the compensation filter is the inverse of the X-filter frequency response or the inverse of the secondary path frequency response.

Preferably, in an alternative embodiment, the frequency response of the compensation filter is the inverse of the frequency response of the secondary path.

Preferably, in an optional embodiment, in this embodiment, the active noise reduction device 10 further includes:

a compensation coefficient calculation module 14, configured to obtain the coefficient of the compensation filter through N iterations; wherein N is an integer.

Preferably, in an optional embodiment, the compensation coefficient calculating module 14 includes:

the adaptive iteration unit is used for keeping the coefficient of the compensation filter unchanged in T time, iteratively updating the coefficient of the adaptive filter through an LMS module, and updating the coefficient of the compensation filter by the coefficient at T moment;

the compensation iteration unit is used for obtaining the coefficient of the compensation filter after the adaptive iteration unit carries out updating for N times; wherein T is a preset convergence time.

Please refer to the description of the foregoing embodiments, and details of the process of the active noise reduction method implemented by the active noise reduction apparatus 10 will not be repeated herein.

The present invention further provides an active noise reduction apparatus, applied in an active noise reduction system, where the active noise reduction apparatus 20 is capable of implementing the active noise reduction method of the foregoing embodiment, as shown in fig. 14, in an embodiment, the active noise reduction apparatus includes an adaptive filter, an X-filter, a compensation filter, and a coefficient updating module, where an input end of the adaptive filter is connected to a reference microphone to receive a reference noise signal X (n), and an output end of the adaptive filter outputs a secondary noise y (n) to a noise cancellation speaker; the input end of the X-filter is connected with the reference microphone to receive the reference noise signal X (n), and the output end of the X-filter is connected with one input end of the coefficient updating module; the input end of the compensation filter is connected with the error microphone to obtain an error noise signal e (n), and the output end of the compensation filter is connected with one input end of the coefficient updating module; the coefficient updating module calculates the coefficient of the adaptive filter and outputs the coefficient to the adaptive filter to update the coefficient of the adaptive filter.

It will be appreciated that in the process from the output of the adaptive filter to the error microphone 22, the driving signal output by the adaptive filter generally needs to undergo DAC conversion, low pass filter, and power amplification to drive the noise cancellation speaker to emit secondary noise that can cancel the reference noise signal. Similarly, after the reference microphone collects noise from the noise source, a reference noise signal (not shown in the figure) can be obtained only by the pre-amplifier, the low-pass filter and the ADC conversion, and after the error microphone collects an error noise signal, the error noise signal also needs to be input into the active noise reduction device as a feedback signal after being converted by the pre-amplifier, the low-pass filter and the ADC, so as to adjust the coefficient of the adaptive filter and further optimize the active noise reduction effect. The embodiment of the invention does not limit the signal preprocessing process before the reference noise signal is acquired and the signal processing process from the driving signal to the secondary noise.

The present invention further provides an active noise reduction system, as shown in fig. 15, in an embodiment, the active noise reduction system 200 includes a reference microphone 1 for picking up noise source noise, an error microphone 2 for picking up an error noise signal, and an active noise reduction device 10 or an active noise reduction device 20 as in the previous embodiments.

In the present embodiment, the path from the reference microphone 1 to the error microphone 2 is a primary path 100, the output from the active noise reduction device adaptive filter to the error microphone 2, and the path from the error microphone 2 to the active noise reduction device 10 becomes a secondary path 200.

The invention further provides a headset having an ear cup and also having an active noise reduction system as described above, wherein the reference microphone is arranged to pick up a noise signal outside the ear cup and the error microphone is arranged to pick up a noise signal inside the ear cup.

The invention further provides a chip having an integrated circuit thereon, the integrated circuit being designed for implementing the active noise reduction method as described in the previous embodiments.

The invention further provides an audio device comprising an active noise reduction means 10 or 20 as described in the previous embodiments.

The present invention further provides a storage medium having stored thereon a computer program which, when executed by a processor, performs the active noise reduction method according to the foregoing embodiments.

It will be appreciated by those skilled in the art that the above-described preferred embodiments may be freely combined, superimposed, without conflict.

It will be understood that the embodiments described above are illustrative only and not restrictive, and that various obvious and equivalent modifications and substitutions for details described herein may be made by those skilled in the art without departing from the basic principles of the invention.